What your doctor is reading on Medscape.com:
MAY 05, 2020 — Medscape has asked top experts to weigh in on the most pressing scientific questions about COVID-19, starting with serology studies. Check out the first in this series by Dr Natalie Dean.
Results of a Los Angeles County antibody study, which was notably not peer reviewed, offer a little insight into the life of an infectious disease epidemiologist trying to handle uncertainty and report responsibly.
Warning: The following discussion will contain many “I’m not exactly sure” statements.
This antibody study is a serology study. We’ve been waiting for good serology for what seems like forever. But it’s still not here. Uncertainties include how well the various tests work, how much immunity antibodies generate, and exactly where a location is on its epidemic curve. So, while the report from Los Angeles suggests that more than 4% of the population has been exposed to the pandemic virus, it remains somewhat hard to know what that means.
First, how certain are we that SARS-CoV-2 antibodies are what was measured? Could the antibodies found instead be a response to one of the milquetoast circulating betacoronas?
This is measured by the test specificity, or the false-positive rate. These rates can be quite high for a lot of serologic tests. The Infectious Diseases Society of America (IDSA) has released a helpful primer. Quoting from it: “Some FDA-authorized COVID-19 antibody tests are estimated to have 96-98% specificity, which would mean that a positive test result is more likely a false-positive result than a true positive result if the prevalence or pretest probability is 5% or less.”
In other words: Be cautious about interpreting serologic studies from places where we don’t think there has been a lot of disease. What they find could easily be false positives, which give a faulty impression of the actual amount of disease—and bad advice to the people tested.
The COVID-19 Testing Project provides comparisons of the performance of the different tests available.
It is also important that a study examining serology take as close to a random sample as possible. Studies doing recruitment via social media are not random, because people may be more likely to participate if they think they have been infected, and want an answer.
I think there’s a quite enormous amount of undetected transmission and immunity (but that is emphatically not the point). The point is what happens when the virus is given free reign. While this has not happened in California, we have seen it in Italy and New York City.
When given free reign, this virus can crash healthcare.
Let’s stop that from happening. We can think later about how many people were infected and never knew. Right now, let’s focus on those who absolutely know they’re infected. New York City has already seen more deaths from this than you would expect for a flu season in which every man, woman, and child was infected. It’s slowing, thank goodness, but not stopping.
Rather than obsessing about the exact number of asymptomatic infections, we need to keep focusing on how to minimize new infections, and refine what we’ve been doing in terms of distancing to help maintain ourselves and our society during the pandemic.
Bill Hanage is an associate professor at the Center for Communicable Disease Dynamics in the Department of Epidemiology at the Harvard T. H. Chan School of Public Health. He specializes in pathogen evolution. Follow him on Twitter